Radio frequency (RF) communications systems, such as cellular and wireless area networks, are ubiquitous in today's world. A key and essential component of every RF communications system is the RF transmitter. As illustrated in FIG. 1, an RF transmitter 100 generally comprises a baseband processor 102, a frequency upconverter 104, a power amplifier (PA) 106 and an antenna 108. The purpose of the baseband processor 102 is to generate a baseband signal s(t) containing a message to be transmitted and formatted in accordance with a predetermined modulation scheme. The purpose of the frequency upconverter 104 is to upconvert the baseband signal s(t) to RF, so that the message is capable of being propagated through space (i.e., transmitted over the air) to a remote receiver. The PA 106 is used to increase the power of the RF signal before it is radiated by the antenna 108, to compensate for attenuation of the RF signal as it is transmitted over the air to the remote receiver.
In modern RF transmitters, the message to be transmitted is first digitized in the form of a binary-source data stream. The baseband processor 102 then groups the data bits in the binary-source data stream into a sequence of N-bit words, where N is some positive integer, and maps the pattern of bits in each N-bit word to one of M=2N possible symbols. The M symbols are defined by the particular modulation scheme being employed, and affect how the amplitude and/or angle of the RF carrier signal is varied (i.e., modulated) to carry the message in the original binary-source data stream to the remote receiver. By mapping each N-bit word to one of M possible symbols, N=log2M bits can be transmitted in each symbol.
Conceptually, the symbols generated by the baseband processor 102 can be visualized as a sequence of weighted impulses. These impulses have essentially infinite bandwidth. To limit their bandwidth, the baseband processor 102 is further configured to shape each symbol by a band-limiting pulse p(t).
Mathematically, the baseband signal s(t) can be expressed as a sequence of pulse-shaped symbols:
            s      ⁡              (        t        )              =                  ∑        n            ⁢                        a          n                ⁢                  p          ⁡                      (                          t              -                              nT                s                                      )                                ,where n is a symbol index, an is the nth symbol in the sequence of symbols, p(t) is the pulse at time t, and Ts is the symbol period. an is either a real or complex number having one of M possible states. For example, in the quadrature phase-shift keying (QPSK) modulation scheme, M=4, and an is given by an=ejπ(2dn+1)/2, where dn is an integer selected from the set {0, 1, 2, 3}.
Because the baseband signal s(t) is in general a complex signal, it is usually expressed in terms of its in-phase (I) and quadrature (Q) components, i.e., as s(t)=I(t)+jQ(t) and the baseband processor 102 is configured to generate separate pulse-shaped I and Q baseband signals for each of the I and Q channels of the RF transmitter.
FIG. 2 is a drawing showing how the baseband signal s(t) is processed in terms of its I and Q components in a practical RF transmitter 200. The RF transmitter 200 comprises a baseband processor 202, I-channel and Q-channel digital to analog converters 204 and 206, a transmit local oscillator (Tx-LO) 208, a quadrature modulator 210; a PA 212; and an antenna 214. Because of its use of the quadrature modulator 210, the RF transmitter 200 is referred to in the description that follows as the “quadrature-modulator-based” transmitter 200.
The quadrature modulator 210 includes an I-channel mixer 216, a Q-channel mixer 218, a ninety-degree phase shifter 220, and a subtractor 222. The I-channel and Q-channel digital to analog converters 204 and 206 convert the pulse-shaped I and Q baseband signals from the baseband processor 202 into analog I and Q baseband signals. The quadrature modulator 210 then upconverts the analog I and Q baseband signals to RF. Specifically, the I-channel mixer 216 mixes the analog I baseband signal with an RF carrier signal provided by the Tx-LO 208, while the Q-channel mixer 218 mixes the analog Q baseband signal with a ninety-degree phase-shifted version of the RF carrier signal produced at the output of the ninety-degree phase shifter 220. The upconverted I- and Q-channel RF carrier signals are then combined by the subtractor 222, to produce the desired modulated RF carrier signal. Finally, the modulated RF carrier signal is amplified by the PA 212 and radiated over the air to a remote receiver by the antenna 214.
One advantage of the quadrature-modulator-based RF transmitter 200 is that both amplitude and angle (i.e., frequency or phase) modulation can be introduced into the RF carrier signal by simply controlling the amplitudes of the I and Q baseband signals. However, a significant drawback is that it is not very energy efficient, particularly when the modulation scheme being employed is a non-constant envelope modulation scheme.
In an effort to use the RF spectrum as efficiently as possible, modern communications systems commonly employ non-constant envelope modulation schemes in which both the amplitude and angle of the baseband signal s(t) are varied. As illustrated in FIG. 3, use of a non-constant envelope modulation scheme results in a modulated RF carrier signal at the RF input RFin of the PA 212 having a non-constant (i.e., time varying) envelope. To prevent the PA 212 from clipping the signal peaks of these signals, the input power of the modulated RF carrier signal must be backed off to ensure that the PA 212 always operates in its linear region of operation. In other words, the PA 212 must be operated as a “linear” PA when a quadrature modulator is used.
While employing power back-off does help to ensure PA linearity, it also results in a significant reduction in energy efficiency. The energy efficiency of an RF transmitter is determined in large part by the efficiency of the RF transmitter's PA. The energy efficiency of the PA is defined as the ratio of the PA RF output power to the direct current (DC) power supplied to the PA 212 from the RF transmitter's constant voltage supply Vs. Energy efficiency is therefore high when the PA is operating at high RF output powers, but low when the PA is operating at low RF output powers. In most applications, the PA operates at high or peak RF output powers only for very short periods of time. For all other times (i.e., most of the time), the RF output power is backed off, resulting in a substantial reduction in energy efficiency.
Low energy efficiency is undesirable in most applications. It is particularly undesirable in battery-powered RF transmitters, such as those used in cellular handsets, since it results in shortened battery life. Fortunately, an alternative type of communications transmitter known as a polar transmitter is available which avoids the linearity versus efficiency tradeoff of the quadrature-modulator-based transmitter 200. In a polar transmitter the amplitude information (i.e., the signal envelope) is temporarily removed from the non-constant envelope signal. The remaining signal, which has a constant envelope, is upconverted to RF and applied to the RF input port of the PA while the previously removed signal envelope is used to dynamically control the power supplied to the PA. Because the signal applied to the RF input of the PA has a constant envelope, a more efficient nonlinear PA can be used without the risk of signal peak clipping.
FIG. 4 is a drawing showing the basic elements of a polar transmitter 400. The polar transmitter 400 comprises a baseband processor 402; a Coordinate Rotation Digital Computer (CORDIC) converter (i.e., rectangular-to-polar converter) 404; an amplitude path including an amplitude path DAC 406 and amplitude modulator 408; an angle path including an angle path DAC 410 and angle modulator 412; a PA 414; and an antenna 416. The purpose of the CORDIC converter 404 is to convert the digital rectangular-coordinate pulse-shaped I and Q baseband signals from the baseband processor 402 to digital polar-coordinate amplitude and angle component signals ρ and θ. The amplitude and angle path DACs 406 and 410 convert the digital amplitude and angle component signals ρ and θ into analog amplitude and angle modulation signals. In the amplitude path, the amplitude modulator 408 then modulates a direct current power supply voltage Vsupply (e.g., as provided by a battery) by the amplitude information in the analog amplitude modulation signal. The resulting amplitude-modulated power supply signal Vs(t) is coupled to the power supply port of the PA 414. Meanwhile, in the angle path the angle modulator 412 operates to modulate an RF carrier signal by the angle information in the analog angle modulation signal, to produce an angle-modulated RF carrier signal which is coupled to the RF input port RFin of the PA 414.
The angle-modulated RF carrier signal at the RF input port RFin of the PA 414 has a constant envelope (see FIG. 5). As alluded to above, this permits the PA 414 to be configured to operate in its nonlinear region of operation (i.e., as a “nonlinear” PA) without the risk of signal peak clipping. Typically the PA 414 is implemented as a highly-efficient switch-mode PA (e.g., as a Class D, E or F switch-mode PA) operating between compressed and cut-off states. When configured in this manner, the envelope information in the amplitude-modulated power supply signal Vs(t) is restored at the RF output RFout of the PA 414 as the PA 414 amplifies the angle-modulated RF carrier signal. By operating the PA 414 as a switch and dynamically controlling the power supplied to it, the polar transmitter 400 is able to achieve significantly higher energy efficiencies than the quadrature-modulator-based RF transmitter 200.
Although the polar transmitter 400 is more energy efficient than the quadrature-modulator-based transmitter 200, the amplitude and angle component signals ρ and θ typically have much higher signal bandwidths than the rectangular-coordinate I and Q baseband signals from which they derive. This so-called “bandwidth expansion” phenomenon occurs during the rectangular-to-polar conversion process performed by the CORDIC converter 404. The high signal bandwidths are manifested as high-frequency events in the amplitude and angle component signals ρ and θ and are highly undesirable. Not only do the high-frequency events tend to degrade the modulation accuracy of the polar transmitter 400, they also cause the transmission spectrum to extend beyond its intended band-limited channel, resulting in adjacent channel interference and an increase in receive band noise. These effects can be very difficult to deal with, especially when modulation accuracy and noise limit standards must be adhered to.
The extent to which high-frequency events appear in the amplitude and angle component signals ρ and θ is very much dependent on the modulation scheme being employed. Modulation schemes that produce signals having a high average-to-minimum power ratio (AMPR) generally have a very large angle component bandwidth. In fact, for modulation schemes that produce signal magnitudes that pass through zero, as illustrated in the signal trajectory diagram in FIG. 6, the signal phase changes very abruptly, by as much as 180 degrees, resulting in an angle component signal θ having essentially infinite bandwidth. Signals of such high bandwidth cannot be accurately processed and transmitted by the polar transmitter 400, or by any type of transmitter for that matter.
Various techniques have been proposed to reduce high-frequency events in polar domain signals. One approach, known as “hole blowing,” involves identifying symbols (or samples of symbols) in the baseband signal s(t) during which the magnitude of the signal falls below a predetermined low-magnitude threshold, and then raising the magnitude of the baseband signal s(t) in the temporal vicinity of the identified symbols or samples so that the AMPR of the signal is reduced. The term “hole blowing” is used since the effect of applying the technique is to produce a “hole” in the signal trajectory diagram of the baseband signal s(t). As illustrated in FIG. 7, the “hole” forces the trajectory of the modified baseband signal ŝ(t) to not pass too close to the origin, resulting in a desired reduction in the bandwidth of the signal.
The conventional hole blowing technique is described in detail in U.S. Pat. No. 7,054,385. As explained there, the baseband signal s(t) is modified by adding correction pulses to it, to form the modified baseband signal:
                    s        ^            ⁡              (        t        )              =                            ∑          n                ⁢                              a            n                    ⁢                      p            ⁡                          (                              t                -                                  nT                  s                                            )                                          +                        ∑          m                ⁢                              b            m                    ⁢                      r            ⁡                          (                              t                -                                  t                  m                                            )                                            ,where r(t) is the correction pulse, m is the perturbation index, tm represents the times when the baseband signal s(t) is perturbed (i.e., the times when the correction pulse r(t) is inserted), and bm is a perturbation sequence representing the amplitude scaling and/or angle shifting applied to the correction pulse r(t).
As shown in FIG. 8, in generating the modified baseband signal ŝ(t) the baseband signal s(t) from the baseband processor 102 is fed forward to an analyzer 802. The analyzer 802 then determines the perturbation times tm by detecting low-magnitude events in the baseband signal s(t) that are below the predetermined low-magnitude threshold. In response to detected low-magnitude events, the analyzer 802 generates the perturbation sequence bm. A pulse-shaping filter 804 generates the correction pulse r(t), scales the pulse by the perturbation sequence bm, and finally adds the scaled correction pulses to the original baseband signal s(t) to produce the desired AMPR-reduced modified baseband signal ŝ(t).
While the conventional hole blowing technique can be helpful in reducing AMPR in some applications, it can be ineffective, deficient or even detrimental in others. The conventional hole blowing technique estimates the signal trajectory of the baseband signal s(t) and identifies low-magnitude events based on only two data points of the baseband signal s(t) at a time. Using only two data points can lead to errors in detecting low-magnitude events and can underestimate the level of AMPR reduction that needs to be performed in order to satisfy a particular design requirement or standards specification. The conventional hole blowing technique also requires that the angle of each correction pulse insertion vector be orthogonal to the signal trajectory of the baseband signal s(t). This rigid requirement on the required angle of the pulse insertion vector can increase the complexity of AMPR-reducing circuitry, and lacks any flexibility that would allow the angle of the pulse insertion vector to be adjusted to satisfy a desired balance or combination of in-band and out-of-band noise performance characteristics. Finally, the conventional hole blowing technique is incapable of taking into account prior modifications to the baseband signal s(t). In some circumstances, this can diminish the overall effectiveness of AMPR reduction. In particular, for those modulation schemes having multiple constellation points near the origin in the complex signal plane, the inability to take into account prior modifications to the baseband signal s(t) can result in the generation of pulse insertion vectors that effectively cancel one another out. Such a result defeats the purpose of hole blowing since the signal trajectory of the baseband s(t) can still pass near or through the origin even after the AMPR reduction process has been applied.
It would be desirable, therefore, to have AMPR-reducing methods and apparatus for communications transmitters that are effective at reducing the AMPR of communications signals but which are not plagued by the drawbacks and limitations associated with conventional hole blowing techniques.